Granite Embedding Multilingual R2: Open Multilingual Embedding Model with 32K Context

Granite Embedding Multilingual R2：开源多语言嵌入模型，具备32K上下文与领先检索性能

IBM Granite released Granite Embedding Multilingual R2 on Hugging Face under Apache 2.0, with fewer than 100 million parameters, a 32K-token context length, and top same-scale retrieval performance on MTEB according to the post.

Why it matters: HKR-H/K/R pass: the 32K-context, sub-100M multilingual embedding model gives RAG builders a concrete open-source option. Impact is narrower than a frontier-model release, so it sits at the featured threshold.